International Journal of Mining Science and Technology xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

International Journal of Mining Science and Technology

journal homepage: www.elsevier.com/locate/ijmst

A discrete-event model to simulate the effect of truck bunching due

to payload variance on cycle time, hauled mine materials and fuel
consumption
A. Soofastaei ⇑, S.M. Aminossadati, M.S. Kizil, P. Knights
School of Mechanical and Mining Engineering, The University of Queensland, Brisbane, QLD 4072, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Data collected from truck payload management systems at various surface mines shows that the payload
Received 29 November 2015 variance is significant and must be considered in analysing the mine productivity, energy consumption,
Received in revised form 13 January 2016 greenhouse gas emissions and associated cost. Payload variance causes significant differences in gross
Accepted 28 March 2016
vehicle weights. Heavily loaded trucks travel slower up ramps than lightly loaded trucks. Faster trucks
Available online xxxx
are slowed by the presence of slower trucks, resulting in ‘bunching’, production losses and increasing fuel
consumptions. This paper simulates the truck bunching phenomena in large surface mines to improve
Keywords:
truck and shovel systems’ efficiency and minimise fuel consumption. The study concentrated on complet-
Discrete-event model
Simulation
ing a practical simulation model based on a discrete event method which is most commonly used in this
Truck bunching field of research in other industries. The simulation model has been validated by a dataset collected from
Payload variance a large surface mine in Arizona state, USA. The results have shown that there is a good agreement
Cycle time between the actual and estimated values of investigated parameters.
Fuel consumption Ó 2016 Published by Elsevier B.V. on behalf of China University of Mining & Technology.

1. Introduction
Improving the efficiency of haulage systems is one of the great
challenges in mining engineering and is the subject of many
research projects undertaken in both study and industry [1–9].
For mining, it is important that haulage systems are designed to
be as efficient as possible, in order to minimise haulage cost,
improve profitability and increase the total mine value. Haulage
system inefficiency is typically derived from inadequate engineer-
ing, which results in poor haul road design, machinery standby and
downtime, and circuit traffic [10–12]. According to the literature,
haulage costs can be some of the largest in a mining system
[13,14]. In various case studies it was found that material trans-
portation represents 50% of the operating costs of a surface mine
[15].
The main effective parameters on material transport when a
truck and shovel system is used in surface mines are mine plan-
ning, road condition, truck and shovel matching, swell factors, sho-
vel and truck driver’s ability, weather condition, payload
distribution and payload variance [16–19]. Based on the literature
among all above mentioned parameters, truck payload variance is
one of the most important parameters in this field [7,20,21]. The
⇑ Corresponding author. Tel.: +61 7 33658232.
E-mail address: a.soofastaei@uq.edu.au (A. Soofastaei).
payload variance not only affects the production rate, but also it
is an important parameter in the analysis of fuel consumption.
The main source of the payload variance in truck and shovel mine
operation is the loading process. Loading is a stochastic process
and excavator performance is dependent on factors such as swell
factor, material density and particle size distribution [22]. Varia-
tion of these factors causes variation of bucket and consequently
truck payloads, affecting productivity. Reducing truck payload
variance in surface mining operations improves productivity by
reducing bunching effects and machine wear from overloaded
trucks [23]. In large surface mines having long ramps, bi-
directional traffic and restrictions on haul road widths negate the
possibility of overtaking. Overloaded trucks are slower up ramp
in comparison to under-loaded trucks. Thus, faster trucks can be
delayed behind slower trucks in a phenomenon known as truck
bunching [20]. This is a source of considerable productivity loss
for truck haulage systems in large surface mines.
There are some investigations about the payload variance sim-
ulation and the effect of this event on other mining operational
parameters. A project completed by Hewavisenthi, is about using
a Monte-Carlo simulation to investigate the effect of bulk density,
fill factor, bucket size and number of loading passes on the long
term payload distribution of earthmoving systems [21]. The focus
of their study is on simulation of payload distribution and variance
in large surface mines. A study conducted by Knights and Paton

http://dx.doi.org/10.1016/j.ijmst.2016.05.047
2095-2686/Ó 2016 Published by Elsevier B.V. on behalf of China University of Mining & Technology.

Please cite this article in press as: Soofastaei A et al. A discrete-event model to simulate the effect of truck bunching due to payload variance on cycle time,
hauled mine materials and fuel consumption. Int J Min Sci Technol (2016), http://dx.doi.org/10.1016/j.ijmst.2016.05.047
2 A. Soofastaei et al. / International Journal of Mining Science and Technology xxx (2016) xxx–xxx
concerned with truck bunching due to load variance [20]. This
study was conducted to provide an analysis of the effect of load
variance on truck bunching. In this project, a GPSS/H model was
constructed which simulates a haulage circuit designed using data
inputs from a real mine site. The model was used to run haul cir-
cuit simulations with different levels of payload variance. From
empirical data, haul route travel times were estimated to be
dependent on payload based on a linear relationship with an addi-
tional stochastic component modelled by a normal distribution.
The data was insufficient to determine the dependence of changes
in haul route travel time on changes in payload variance. In this
project, a simulation was also conducted to investigate the haul
circuit throughput difference if single truck overtaking was permit-
ted. Webb investigated the effect that different bucket load sizes
had on truck cycle times and the inherent costs [24]. The research
project being undertaken will focus primarily on the effect of load
variance on truck bunching.
Based on the condition of truck and shovel mining operations in
surface mines, the best simulation for this event can be simulated
by discrete event methods. Discrete event simulation can be used
to model systems which exhibit changes in state variables at a dis-
crete set of points in time [26]. The models can be static or
dynamic. Static models represent a system at a specific time, while
dynamic models represent a system as it evolves over a period of
time [26]. A mining operation is a dynamic system which is very
difficult to model using analytical methods. When simulation is
used, the model input can be based on probabilistic data which
better characterise the input variables and a given number of vari-
ables can be described by selecting appropriate distributions [27].
The trucks utilised in the haulage operations of surface mines
consume a great amount of fuel and this has encouraged truck
manufacturers and major mining corporations to carry out a num-
ber of research projects on the fuel consumption of haul trucks
[28]. There are many factors that affect the rate of fuel consump-
tion for haul trucks such as payload, velocity of truck, haul road
condition, road design, traffic layout, fuel quality, weather condi-
tions and driver skill [1]. A review of the literature indicates that
understanding of energy efficiency of a haul truck is not limited
to the analysis of vehicle-specific parameters; and mining compa-
nies can often find greater energy saving opportunities by expand-
ing the analysis to include other effective factors such as payload
distribution and payload variance [29].
This paper aims to present a new simulation model based on
the discrete event methods to investigate the effect of truck bunch-
ing due to payload variance on average cycle times, the rate of
loading materials and fuel consumption.
2. Payload variance
Loading performance depends on different factors such as
bench geometry, blast design, muck pile fragmentation, operators’
efficiency, weather conditions, utilisation of trucks and shovels,
mine planning and mine equipment selection [21,30]. In addition,
for loading a truck in an effective manner, the shovel operator must
also strive to load the truck with an optimal payload. The optimal
payload can be defined in different ways, but it is always designed
so that the haul truck will carry the greatest amount of material
with lowest payload variance [20]. The payload variance can be
illustrated by carrying a different amount of overburden or ore
by the same trucks in each cycle. The range of payload variance
can be defined based on the capacity and power of the truck. The
increase of payload variance decreases the accuracy of the mainte-
nance program. This is because the rate of equipment wear and
tear is not predictable when the mine fleet faces a large payload
variance [23]. Minimising the variation of particle size distribution,
Please cite this article in press as: Soofastaei A et al. A discrete-event model to simulate the effect of truck bunching due to payload variance on cycle time,
hauled mine materials and fuel consumption. Int J Min Sci Technol (2016), http://dx.doi.org/10.1016/j.ijmst.2016.05.047
swell factors, material density and fill factor can decrease the pay-
load variance but it must be noted that some of the mentioned
parameters are not controllable. Hence, the pertinent methods to
minimise the payload variance are real-time truck and shovel pay-
load measurement, better fragmentation through optimised blast-
ing and improvement of truck-shovel matching. The payload
variance can be shown by variance of standard deviation r. Stan-
dard deviation measures the amount of variation from the average.
A low standard deviation indicates that the data points tend to be
very close to the mean; a high standard deviation indicates that the
data points are spread out over a large range of values. This param-
eter can be calculated by
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 XZ
r¼ i¼1 i
ðx  lÞ2 ð1Þ
Z
where Z is the number of available data for each parameter; i the
counter of data; x the value of parameter; and l the mean which
can be calculated by following equation.
1X Z
l¼ xi ð2Þ
Z i¼1
Fig. 1 shows the different kinds of normal payload distribution
(the best estimation function for payload distribution) based on
the difference r for one type of the mostly used truck in surface
mines (CAT 793D).
In Fig. 1, gross vehicle weight (GVW) is the total weight of
empty truck and payload. Based on the CAT 793D technical speci-
fications, the range of GVW variation is between 165 (empty truck)
and 385 tonnes (maximum payload). Hence, the maximum r for
this truck can be defined as 30; that is because for higher standard
deviations, the minimum GVW is less than the weight of empty
truck and the maximum GVW is more than the maximum capacity
of truck.
3. Discrete simulation modelling
Based on the condition of truck and shovel mining operation in
surface mines, the best simulation for this event can be by discrete
event methods. Discrete event simulation can be used to model
systems which exhibit changes in state variables at a discrete set
of points in time [25,31]. The models can be static or dynamic. Sta-
tic models represent a system at a specific time, while dynamic
models represent a system as it evolves over a period of time
[32]. A mining operation is a dynamic system which is very diffi-
cult to model using analytical methods. There are different kinds
of discrete simulation models used for modelling the systems in
industrial projects. In this study, some of the most popular models
have been investigated and a new model to simulate the truck
bunching event in surface mining operation has been developed.
The first investigated model is AutoMod. This model is a simu-
lation system which is designed for use in material movement sys-
tems developed by Applied Materials, USA [33]. It can be used for
80 σ =5
σ = 10
Gross vehicle weight frequency
70
σ = 15
60
σ = 20
50 σ = 25
σ = 30
40
30
20
10
0
150 175 200 225 250 275 300 325 350 375 400
Gross vehicle weight (tonne)
Fig. 1. Normal payload distribution for different standard deviation r (CAT 793D).
 A. Soofastaei et al. / International Journal of Mining Science and Technology xxx (2016) xxx–xxx
simulation of truck haulage circuits and transport circuits, convey-
ors, load dumping and retrieval, cranes and robots. Simulations
with AutoMod have the ability for simulation of complex move-
ment with stochastic inputs. AutoMod models can contain multi-
ple systems (e.g. interacting truck and shovel circuits). To
produce a simulation, the user constructs a series of action state-
ments which allow the incorporation of elements such as machin-
ery, queues, loading, delays and input values/variables.
Simulations also allow the use of conditional tests. Load inputs
can be deterministic or stochastic. AutoMod offers control vari-
ables for queuing, wait times and traffic which are crucial for haul
circuit simulation bunching analysis. Visualisation of simulation is
powerful and extensive in AutoMod. Graphical model outputs can
be represented in three dimensions and is industry leading in
terms of animation and realism.
The second studied model is SIMUL8. This model is a graphically
oriented simulation package developed by the SIMUL8 Corporation
[34]. This software is a discrete event simulation package, meaning
it simply executes tasks in queue based on time, which then trig-
gers the activity of new tasks. SIMUL8 can be used in simulation
of multiple haulage systems, but is more effective at single circuit
simulations.
The third analysed model is GPSS/H. The general purpose simu-
lation system (GPSS) language was originally released in 1961 and
became a popular means of simulation since it could be operated
without the requirement for the user to be knowledgeable in pro-
gramming. GPSS/H was derived from the evolution and expansion
of GPSS and became the more widespread and superior package.
GPSS/H was released in 1977 by Wolverine Software Corporation
who still develops and sells GPSS/H today [20]. GPSS/H can be used
with a wide range on models due to its simplicity and flexibility. It
is based on a flowchart type system using ‘‘transactions” which
move between ‘‘blocks”. It involves the creation of blocks and con-
trols statements to generate a system. Transactions move through-
out the system based on the tick of an internal clock. Each tick of
the clock corresponds to one-time unit worth of action. GPSS/H
is stochastic in nature, such that it can execute Monte Carlo style
randomisation to apply statistical distributions. GPSS/H is particu-
larly adept at simulating queuing and bunching. GPPS/H can be
applied to several systems including haulage circuits, data flow
or a production line. The language is based on text entry, and does
not provide visualisation without the use of proof animation.
The fourth studied model in this project is WITNESS. This model
is a discrete event simulation suite developed by Lanner. Witness is
capable of producing haulage system simulations in a dynamic ani-
mated computer model [35]. The suite consists of four separate
modules, the main WITNESS simulation module, an experimenta-
tion optimiser, a scenario manager for analysis and a three dimen-
sional visual output.
The last but not least inspected model is Arena. This model is a
simulation software package developed by Rockwell Automation
based on the SIMAN programming language [36]. SIMAN is a dis-
crete event simulation package which can be used in process or
event scheduling mode. SIMAN is most commonly used in conjunc-
tion with Arena in industry today. SIMAN can alternatively be used
in conjunction with CINEMA, a visualisation package. The ARENA
system can produce scale models of circuits and other simulations.
4. Truck bunching model
4.1. Developed algorithm
Hauling operations in surface mines consists of different kinds
of components. These components are loading, hauling, manoeu-
vring, dumping, returning and spotting (Fig. 2).
Please cite this article in press as: Soofastaei A et al. A discrete-event model to simulate the effect of truck bunching due to payload variance on cycle time,
hauled mine materials and fuel consumption. Int J Min Sci Technol (2016), http://dx.doi.org/10.1016/j.ijmst.2016.05.047
3
In the standard hauling operation, loading time is the time
taken to load the truck, and hauling and returning time are travel-
ling time for each truck between loading zone and dumping area.
Spotting time is the time during which the loading unit has the
bucket in place to dump, but is waiting for the truck to move into
position. Spotting time will depend on the truck driver’s ability and
the loading system. Double-side loading should almost eliminate
spot time. Dumping time is the time taken for the truck to
manoeuvre and dump its payload either at a crusher or dump.
Based on the above mentioned hauling operation components,
four main times can be defined; fixed time, travel time, wait time
and cycle time.
Fixed time is sum of the loading, manoeuvring, dumping and
spotting time. It is called ‘fixed’ because it is essentially invariable
for a truck and loading unit combination. Travel time is the time
taken to haul and return the payload. Wait time is the time the
truck must wait before being served by the loading unit, waiting
in a queue for dumping and the waiting time in line behind the
overloaded trucks in large surface mines (truck bunching). Cycle
time is the round trip time for the truck. It is the sum of the fixed,
travel and wait times.
Fig. 3 illustrates the proposed algorithm to complete a discrete
event model in this project.
This algorithm consists of four main subroutines to cover all
processes in the hauling operation. These main components are
loading, hauling, dumping and returning. Based on the developed
model, each component has a waiting time. The main reason for
waiting time in hauling is payload variance.
4.2. Payload distribution and variance simulation
A main part of the truck bunching model is simulating the pay-
load distribution and variance. In this study, a simulation model
was designed to estimate the distribution of truck and bucket pay-
loads based on several of input parameters. These parameters are
bucket size, number of loader passes (to fill the truck tray), distri-
bution of bucket bulk density and distribution of bucket fill factor.
This simulation was implemented as a MATLAB workbook and a
commercially available Monte-Carlo simulation engine was used
to run the simulation. In this model, the truck payload is calculated
by
X
P
mk ¼ qk v bf q ð3Þ
q¼1
where mk is the truck payload (for the kth truck); Vb the bucket
rated capacity; fq the fill factor; qk the bucket density (one value
for all of the passes in one truck); q the bucket pass; and P the max-
imum bucket passes to fill the truck tray. In this simulation bucket
bulk density (qk) and fill factor (fq) are randomly selected by the
Monte-Carlo simulation engine.
Loading Hauling
Manoeuvring
Spotting
Dumping
Returning
Fig. 2. Schematic of hauling operation in surface mines.
4 A. Soofastaei et al. / International Journal of Mining Science and Technology xxx (2016) xxx–xxx

Loading Hauling Dumping Returning

Yes
Face loading

Is truck in the
Is truck in the Yes first segment ?
Is dumping zone
Is shovel free last segment ? free ?
No
No
No
No
Yes Yes
Is next segment
Spotting Is next segment Manoeuvring occupied ?
occupied ? Yes
Yes
No No

Go to the next Dumping Go to the next

Loading
segment segment

Truck queue Truck queue Truck queue Truck queue

Queuing for Queuing in Queuing for Queuing in

loading hauling dumping returning

Fig. 3. Truck bunching algorithm.

4.3. Model considerations ðTravel timeÞi;k ¼ t ðTÞi;k

X 2li ðV ðiþ1Þ;ðk1Þ  V ði1Þ;k Þ
In the model, the total length of haul and return road is divided ¼ ni;k
in segments based on the variation of total resistance (TR). TR is i V 2ðiþ1Þ;ðk1Þ  V 2ði1Þ;k
 
equal to the sum of the rolling resistance (RR) and grade resistance 2li V i;k  V ði1Þ;k
(GR). The haul and return road are analysed using the same þ ð1  ni;k Þ ð8Þ
V 2i;k  V 2ði1Þ;k
approach. However, on haul roads, the grade resistance is positive
and on the return road it is negative (Fig. 4). The main reason for where t(T)i,k is the travel time for truck k in segment i; li the length of
truck bunching on haul road is payload variance and the reason segment i; Vi,k the velocity of truck k in segment i; V(i1),k the veloc-
for truck bunching on return roads is the driver’s supposed ability ity of truck k in segment i  1; and ni,k the decision variable.
and mine traffic management. XX
Hauled mine materials ¼ payloadk;r =shift time ð9Þ
r k
4.4. Decision variables
where payloadk,r is the payload of truck k in cycle r.
In completed discrete event model three decision variables have
been defined. The variables are Uk, Sk and ni,k. 4.6. Constraints

1 If Truck k is in first segment There are three main constraints in the presented model.
Uk ¼ ð4Þ
0 Otherwise X
li ¼ 2L ¼ Lengh of haul road þ lenght of return road ð10Þ
 i
1 If Truck k is in last segment
Sk ¼ ð5Þ
0 Otherwise

1 If V i;k > V i;ðk1Þ
ni;k ¼ ð6Þ Dumping
0 Otherwise
Re
tur
To create a practical model, it is necessary to define some func- ng nin
uli g
tions based on the above mentioned decision variables. Ha

4.5. Objective functions GR(+) GR(0) GR(-)

Gradient Gradient
In this section, the objective functions for cycle time, travel time
and hauled mine materials have been presented in following
equations. Fig. 4. Grade resistance (GR).
X
ðCycle timeÞk ¼ t S þ t L þ t ðTÞi þ t M þ tD þ ðt S þ t L ÞW okj U k
i

þ ðt M þ tD ÞW Lkj Sk ð7Þ Table 1

A simplified version of the payload matrix (Pk,r).
where ts is the spotting time; tL the loading time; tT the travel time;
tM the manoeuvring time; tD the dumping time; W0kj the number of Cycle (r) Truck (k) k=1 k=2 ... k=N

trucks at queue in front of truck k at time j in the first segment; WLkj r=1 ? P1,1 P2,1 ... PN,1
the number of trucks at queue in front of truck k at time j in the last r=2 ? P1,2 P2,2 ... PN,2
... ... ... ... ... ...
segment; Uk the first decision variable; and Sk the second decision
r=M ? P1,Z P2,Z ... PN,Z
variable.

Please cite this article in press as: Soofastaei A et al. A discrete-event model to simulate the effect of truck bunching due to payload variance on cycle time,
hauled mine materials and fuel consumption. Int J Min Sci Technol (2016), http://dx.doi.org/10.1016/j.ijmst.2016.05.047
 A. Soofastaei et al. / International Journal of Mining Science and Technology xxx (2016) xxx–xxx 5

Table 2 4.7. Data processing

A simplified version of velocity matrix (Pk,i, Vki & t(T)k,i).

Segment (i) Truck (k) k=1 k=2 k=N The developed truck bunching model uses two matrices at the
i=1 ? P1,1 P2,1 ... PN,1 same time (parallel processing) to create and process data. The first
V1,1 V2,1 ... VN,1 matrix is used to generate the truck payload based on Eq. (3). In
t(T)1,1 t(T)2,1 ... t(T)N,1 this process the truck payload in all steps of the model will be gen-
i=2 ? P1,2 P2,1 ... PN,2 erated randomly by a Monte-Carlo simulation engine. A simplified
V1,2 V2,1 ... VN,2 version of payload matrix is presented in Table 1. In Table 1, k rep-
t(T)1,2 t(T)2,1 ... t(T)N,2 resents the number of trucks and r represents the number of cycles
... ... ... ... ... ... in each shift; and Pk,r is the payload of truck k in cycle r.
... ... ... ...
... ... ... ...
i = 2L ? P1,2L P2,1 ... PN,2L
V1,2L V2,1 ... VN,2L
t(T)1,2L t(T)2,1 ... t(T)N,2L Table 5
A sample of real mine site parameters (a case study).

Parameter Value

Table 3 (a) Material

Data collected for model validation (sample). Insitu bank density (tonne/m3) 2.5

No. Average loader Truck Average bucket Loader Average Swell factors (tonne/m3) Bank to loader bucket 1.25
payload payload bulk density bucket fill swell Bank to loader bucket 1.25
(tonne/pass) (tonne) (tonne/m3) factor factor Lose density (tonne/m3) Bank to truck tray 2
Bank to loader bucket 2
1 47.23 218.21 2.01 0.937 1.25
2 45.12 217.46 1.98 0.978 1.22 Product ration (tonne of product per tonne hauled) 1
3 38.14 209.42 1.96 0.919 1.18
Loader bucket fill factor Heaped 0.978
4 42.15 210.36 2.03 0.954 1.27
Struck 0.978
5 46.58 216.78 2.14 0.984 1.19
6 47.56 217.96 1.86 0.927 1.26 Parameter Value
7 39.87 218.04 2.07 0.946 1.24 (b) Roster for 5 day week-8 h shifts
8 38.47 218.43 2.18 0.992 1.25 Mon–Fri (daily) 3 Shift
9 42.58 217.69 2.05 0.957 1.20 Total shift (shifts/year) 783
10 40.59 216.97 1.99 0.939 1.25 Scheduled lost shifts (shifts/year) 27
Scheduled shifts (shifts/year) 756
Loading unit maintenance (shifts/year) 113
Table 4 Unscheduled lost shift (shifts/year) 42
Sample values of estimated (model) and independent (tests) cycle time and hauled Fleet operating shifts (shifts/year) 601
mine materials. Shift duration (hh:mm:ss) 08:00:00
Non-operating shift delays (hh:mm:ss) 01:00:00
No. Estimated value of cycle Independent value of cycle Absolute In shift operating time (hh:mm:ss) 07:00:00
time (model) (s) time (tests) (s) error (%) Operating shift delays (hh:mm:ss) 00:30:00
In shift working time (hh:mm:ss) 06:30:00
(a) Values of estimated (model) and independent (tests) cycle time (sample)
1 1520 1560 2.56 (c) Loading
2 1650 1680 1.78 Bucket capacity (m3) 25.2
3 1410 1380 2.17 Bucket cycle time (min) 0.5
4 1620 1680 3.57 Mechanical availability 85%
5 1990 2040 2.45 Truck positioning Single sided
6 1910 1860 2.69 Bucket fill factor 0.98
7 1465 1500 2.34 First bucket pass delay (min) 50%
8 1350 1380 2.17 Payload distribution (right skewed) Normal
9 1910 1860 2.69
(d) Truck
10 1390 1440 3.48
Spot time at loader (min) 0.5
Spot time at dump (min) 0.5
No. Estimated average value of Independent average value Absolute Dumping time (min) 0.5
hauled mine materials of Hauled mine materials error (%) Mechanical availability 80%
(model) (tonne/cycle) (tests) (tonne/cycle) Motor power (kW) 1743
Transmission speed factor 1:00
(b) Values of estimated (model) and independent (tests) hauled mine materials
Standard body capacity (m3) 129
(sample)
Empty truck weight (tonne) 165.75
1 203 198 2.46
Actual truck payload (tonne) 218
2 205 200 2.44
Full truck weight (tonne) 383.75
3 207 202 2.42
Operating hours per year 4799.2
4 209 206 1.44
Average payload (tonne) 221.53
5 212 208 1.89
Production per operating hour (tonne) 560.21
6 214 218 1.87
Production per loader operating shift (tonne) 3137.17
7 214 209 2.34
Production per year (tonne) 2688552.13
8 223 228 2.24
Queue time at loader (min/cycle) 2.71
9 229 224 2.18
Spot rime at loader (min/cycle) 0.5
10 233 238 2.15
Average loading time (min/cycle) 1.95
Travel time (min/cycle) 15.94
Spot time at dump (min/cycle) 0.5
ni;k ¼ nk;i ð11Þ Average dump time (min/cycle) 0.5
Average cycle time (min/cycle) 22.11
Fleet size 8
W i;j;k ¼ W i;k;j ð12Þ
Average No. of bucket passes 5

Please cite this article in press as: Soofastaei A et al. A discrete-event model to simulate the effect of truck bunching due to payload variance on cycle time,
hauled mine materials and fuel consumption. Int J Min Sci Technol (2016), http://dx.doi.org/10.1016/j.ijmst.2016.05.047
6 A. Soofastaei et al. / International Journal of Mining Science and Technology xxx (2016) xxx–xxx

The presented model calculates the best performance velocity 28

of each truck in each segment based on the payload generated by 26

Average cycle time (min)

the payload matrix and truck rim pull curve. This model can apply 24
the truck bunching effects on the velocity and hauled mine mate- 22
rial by trucks in each cycle and each segment. A very simplified 20
version of velocity matrix is presented in Table 2. 18
In Table 2, k is the number of trucks in the fleet; i the number of 16
segments in haul and return roads; Pk,i the payload of truck k in
14
segment i; Vk,i the velocity of truck k in segment i; and t(T)k,i the tra-
12
vel time for truck k in segment i. 0 5 10 15 20 25 30 35
The developed parallel data processing in this model can simu- Payload variance standard deviation σ (tonne)
late complicated fleets in large surface mines.
Fig. 6. Variation of average cycle time with payload variance, standard deviation (a
case study).
4.8. Fuel consumption simulation

Haul truck fuel consumption is a function of various parame- 130

ters. The key parameters that affect the fuel consumption of haul
120
trucks include the payload management, the model of the truck,

materials (ktonne/day)
Average hauled mine
the grade resistance and the rolling resistance, according to a study 110
conducted by the Department of Resources, Energy and Tourism
[28]. In the present study, the effects of GVW, the velocity of truck 100
(V) and the TR on the fuel consumption of the haul trucks were
90
examined. The truck fuel consumption can be calculated from Eq.
(13) [37]. 80
0 5 10 15 20 25 30 35
FC ¼ 0:3ðLF  PWÞ ð13Þ
Payload variance standard deviation σ (tonne)

where LF is the engine load factor and is defined as the ratio of aver-
Fig. 7. Variation of average hauled materials with payload variance/standard
age payload to the maximum load in an operating cycle; and PW the deviation (a case study).
truck power, kW [18]. The developed model, in this project, can
simulate the fuel consumption by haul trucks based on Eq. (13).
Table 6
Truck specification (a case study).
4.9. Model validation
Truck specification CAT777F CAT785C CAT793D
To validate the developed model, a dataset collected from a Engine Model C32 3512B 3516B HD
large open pit mine in central Arizona, USA has been applied. This ACERTTM EUI EUI
dataset included measuring average loader payloads, truck pay- Gross power 758 kW 1082 kW 1801 kW
Net power 700 kW 1005 kW 1743 kW
loads, average bucket bulk density, loader bucket fill factor and
average swell factor (Table 3). Weights Total empty operating 64 tonnes 105 165 tonnes
weight tonnes
In this mine, the volume of material loaded into the bucket was
Nominal payload class 96 tonnes 144 218 tonnes
determined by comparing loaded and empty laser scan profiles of tonnes
the buckets. Fill factors were calculated by dividing the material Gross machine operating 160 249 383 tonnes
volume by the rated volume of the bucket and bulk densities were weight tonnes tonnes
calculated by dividing the payload by the loaded volume. On-board
payload monitoring systems were used to measure payloads. The
validation of the model was completed for average cycle times highlight the insignificance of the values of absolute error in the
and the average mine material hauled by one type of truck (CAT analysis.
793D) after truck bunching. Table 4 and Fig. 5 present sample val- The results indicate good agreement between the actual and
ues for the estimated (using the developed model) and the inde- estimated values of average cycle time and average hauled mine
pendent (tested) cycle time and hauled mine material in order to materials.

40 240

35
Estimated value of hauled mine

230
materials (tonne/cycle)
Estimated value of

30
cycle time (min)

2
R =0.95 220 2
25 R =0.90
210
20

15 200

10 190
13 15 17 19 21 23 25 27 29 31 33 35 200 205 210 215 220 225 230 235
Actual value of cycle time (min) Actual value of hauled mine materials
(tonne/cycle)
(a) Comparison of actual values of cycle time with model (b) Comparison of actual values of hauled mine materials
outputs for test data with model outputs for test data

Fig. 5. Comparison of actual values of cycle time and hauled mine materials with model outputs for test data.

Please cite this article in press as: Soofastaei A et al. A discrete-event model to simulate the effect of truck bunching due to payload variance on cycle time,
hauled mine materials and fuel consumption. Int J Min Sci Technol (2016), http://dx.doi.org/10.1016/j.ijmst.2016.05.047
 A. Soofastaei et al. / International Journal of Mining Science and Technology xxx (2016) xxx–xxx 7

Table 7
Fuel consumption index for three models of studied haul truck (a case study) (L/htonne).

Standard deviation r (tonne) CAT 777F CAT 785C CAT 793D

Total resistance (%) TR = 5 % TR = 10% TR = 15% TR = 5% TR = 10% TR = 15% TR = 5% TR = 10% TR = 15%
r = 5 tonnes 0.361 0.459 0.540 0.321 0.399 0.479 0.300 0.375 0.455
r = 10 tonnes 0.456 0.543 0.619 0.416 0.482 0.563 0.399 0.466 0.546
r = 15 tonnes 0.523 0.598 0.671 0.483 0.538 0.618 0.471 0.528 0.608

In Table 7, FCIndex was calculated for three payload standard

0.7
deviations (r = 5, 10 and 15 tonnes) in three different road condi-
0.6 CAT 777F tions (TR = 5, 10 and 15%). The results show that FCIndex increases
0.459 L/h·tonne
not only by increasing the TR but also by increasing the payload
Fuel consumption index (L/h·tonne)

CAT 777F CAT 793D CAT 785C

0.543 L/h·tonne
0.5
CAT 777F
0.459
L/h·tonne
0.4
CAT 793D
0.375
0.3 L/h·tonne
0.2
0.1
0 5
Payload standard deviation σ
Fig. 8. Fuel consumption index for three models of haul trucks, TR = 10% (a case
study).
5. A case study
In this project, a real mine site dataset that was collected from a
large surface mine in central Queensland, Australia has been anal-
ysed. A sample of real mine site parameters is tabulated in Table 5.
Production per year for haulage system is 21,508,417 tonnes.
The effect of truck bunching due to payload variance on average
cycle time and average hauled materials for one mostly used model
of haul truck in studied surface mine is illustrated in Fig. 6.
Fig. 6 demonstrates that, there is a non-linear relationship
between payload variance/standard deviation and average cycle
time in the fleet. Based on the presented results of analysed data
in Fig. 6, it is clear that by increasing the payload variance the aver-
age cycle time increases dramatically. By maximum reducing of
standard deviation from 30 to 5 tonnes, reducing average cycle
time up to 15 min is possible. Other main effective parameters
on mine productivity are average hauled materials. Fig. 7 illus-
trates the relationship between the payload variance/standard
deviation and average hauled materials. The correlation between
mentioned parameters in Fig. 7 is non-linear. The minimum aver-
age hauled mine is obtained with maximum payload variance. The
presented relationship between payload standard deviation and
average hauled materials in studied mine shows that there is a
great opportunity to improve productivity by reducing payload
variance.
In this case study, the effect of payload variance on haul truck
fuel consumption in different haul road conditions for three mod-
els of haul truck has been investigated. It is noted that, to have a
better understanding in this study, a fuel consumption index
(FCIndex) has been defined. This index presents the quantity of fuel
used by a haul truck to move one tonne of mine material (ore or
overburden) in an hour. Truck specifications for studied haul trucks
are presented in Table 6.
Haul trucks were selected based on their capacity and engine
power. The maximum GVW for trucks is 160, 249 and 383 tonnes
respectively. The results of completed investigation by developed
truck bunching model are tabulated in Table 7.
0.528 L/h·tonne 0.538 variance for each truck. Fig. 8 presents the FCIndex versus payload
L/h·tonne
CAT 793D CAT 785C
0.466 L/h·tonne 0.482 standard deviation for three studied models of trucks in same road
CAT 785C L/h·tonne
0.399
condition (TR = 10%).
L/h·tonne Fig. 8 shows that by increasing the capacity of truck, FCIndex can
be reduced. In this case the maximum reduction of FCIndex can be
achieved by changing the model of truck from CAT777F to
CAT 777F
CAT793D.
CAT 785C
CAT 793D
6. Conclusions
10 15 20
This paper aimed to develop a discrete event model to simulate
the effect of payload variance on truck bunching to improve pro-
ductivity and energy efficiency in surface mines. There is a signif-
icant payload variance in the loading process in surface mines. The
main reason for truck bunching in this type of mine is the variance
of payload. In this paper, an innovative simulation model was
developed to investigate the effects of payload variance on truck
bunching, mine operation efficiency and decreasing the fuel con-
sumption by haul trucks. To validate the developed model a data-
set collected from a large surface mine in the central part of
Arizona State, USA was used. Validation of the model was com-
pleted for the cycle time and the hauled mine materials by one
type of truck (CAT 793D) after truck bunching. The results indi-
cated a good agreement between the actual and estimated values
of cycle time and hauled mine materials. The model was utilised
in a real mine site in central Queensland, Australia as a case study.
The results of this project showed that there is a non-linear rela-
tionship between payload variance and cycle time in the fleet. In
this case study, a correlation between the payload variance and
hauled mine materials was developed and the effect of truck
bunching due to payload variance on energy consumption for three
models of haul truck was studied.
Acknowledgments
The authors would like to acknowledge CRC Mining and The
University of Queensland for their financial support for this study.
References
[1] Cetin N. Open-pit truck/shovel haulage system simulation, vol.
1. Ankara: Middle East Technical Universality; 2004. p. 147–56.
[2] Carter T. Failures in a load–haul–dump vehicle axle used in deep mining
operations. Eng Fail Anal 2008;15(7):875–80.
[3] Ghojel J. Haul truck performance prediction in open mining operations. In:
Proceedings of the national conference publication. Australia: Institution of
Engineers; 1993.
[4] Soofastaei A, Aminossadati SM, Arefi MM, Kizil MS. Development of a multi-
layer perceptron artificial neural network model to determine haul trucks
energy consumption. Int J Min Sci Technol 2016;26(2):285–93.
[5] Soofastaei A, Aminossadati SM, Kizil MS. Development of an artificial
intelligence model to determine trucks energy consumption. In: Proceedings
of energy future conference. Future energy, 1. p. 178–9.
[6] Soofastaei A, Aminossadati SM, Kizil MS, Knights P. Payload variance plays a
critical role in the fuel consumption of mining haul trucks. Aust Resour Invest
2014;8(4):64.

Please cite this article in press as: Soofastaei A et al. A discrete-event model to simulate the effect of truck bunching due to payload variance on cycle time,
hauled mine materials and fuel consumption. Int J Min Sci Technol (2016), http://dx.doi.org/10.1016/j.ijmst.2016.05.047
8 A. Soofastaei et al. / International Journal of Mining Science and Technology xxx (2016) xxx–xxx
[7] Soofastaei A, Aminossadati S, Kizil MS, Knights P. Simulation of payload
variance effects on truck bunching to minimise energy consumption and
greenhouse gas emissions. In: Proceedings of the 2015 coal operators’
conference. Australia: The University of Wollongong; 2015.
[8] Guo WB, Xu FY. Numerical simulation of overburden and surface movements
for Wongawilli strip pillar mining. Int J Min Sci Technol 2016;26(1):71–6.
[9] Benton DJ, Iverson SR, Martin LA, Johnson JC, Raffaldi MJ. Volumetric
measurement of rock movement using photogrammetry. Int J Min Sci
Technol 2016;26(1):123–30.
[10] Eskandari H, Darabi H, Hosseinzadeh SA. Simulation and optimization of
haulage system of an open-pit mine. In: Proceedings of simulation series,
37. Society for Modeling & Simulation International; 2013.
[11] Leonardi K. Mining and haulage systems-new challenges and new strategy of
DFM ZANAM-LEGMET Spólka z o.o. J Mines Met Fuels 2008;56(11):223–7.
[12] Cardu M, Lovera E, Patrucco M. Loading and haulage in quarries: criteria for
the selection of excavator–dumper system. In: Proceedings of the 14th
international symposium on mine planning and equipment selection, MPES
2005 & the 5th international conference on computer applications in the
minerals industries. CAMI; 2005.
[13] Benito R, Dessureault SD. Estimation of incremental haulage costs by mining
historical data and their influence in the final pit limit definition. Min Eng
2008;60(10):44–9.
[14] Yuan DC, Yue XG. SVMR model for coal mine cost management prediction.
Appl Mech Mater 2014:608–11.
[15] Curry JA, Ismay MJL, Jameson GJ. Mine operating costs and the potential
impacts of energy and grinding. Miner Eng 2014;56:70–80.
[16] Chironis NP. Coal age operating handbook of coal surface mining and
reclamation. Coal Age Min Inf Serv 1978(2).
[17] Beckman R. Haul trucks in Australian surface mines 2012:87–96.
[18] Kecojevic V, Komljenovic D. Haul truck fuel consumption and CO2 emission
under various engine load conditions. In: SME annual meeting and exhibit,
CMA 113th national western mining conference, USA. p. 186–95.
[19] Fiscor S. Improving haul truck productivity. Coal Age 2007:31–6.
[20] Knights P, Paton S. Payload variance effects on truck bunching. In: Proceedings
of the seventh large open pit mining conference. Melbourne: The Australasian
Institute of Mining and Metallurgy; 2010.
[21] Hewavisenthi R, Lever P, Tadic D. A Monte Carlo simulation for predicting
truck payload distribution. In: Proceedings of the 2011 Australian mining
technology conference, Sydney.
Please cite this article in press as: Soofastaei A et al. A discrete-event model to simulate the effect of truck bunching due to payload variance on cycle time,
hauled mine materials and fuel consumption. Int J Min Sci Technol (2016), http://dx.doi.org/10.1016/j.ijmst.2016.05.047
[22] Singh S, Narendrula R. Productivity indicators for loading equipment. CIM
Magazine; 2006.
[23] Paton S. Truck bunching due to load variance. Australia: The University of
Queensland; 2009.
[24] Webb B. Effects of bucket load distribution on performance. Australia: The
University of Queensland; 2008.
[25] Nelson BL, Carson JS, Banks J. Discrete event system simulation.. USA: Prentice
Hall; 2001.
[26] Basu AJ, Baafi EY. Discrete event simulation of mining systems: current
practice in Australia. Int J Surf Min Reclam Environ 1999;13(2):79–84.
[27] Hu LP, Wang DM, Zuo L. Event step method on computer simulation of discrete
event system. Appl Mech Mater 2014;543:1848–51.
[28] EEO. Analyses of diesel use for mine haul and transport operations. Australian
Government, Energy and Tourism Editor; 2012. p. 1–22.
[29] White JW, Olson JP, Vohnout SI. On improving truck/shovel productivity in
open pit mines. In: Bandopadhyay PS, editor. Proceedings of the 23th
application of computers and operations research in the mineral
industry. Fairbanks: Department of Mining and Geological Engineering
University of Alaska Fairbank; 1992. p. 739–46.
[30] Schexnayder C, Weber S, Brooks B. Effect of truck payload weight on
production. J Constr Eng Manage 1999;125(1):1–7.
[31] Jerry B. Discrete-event system simulation. India: Pearson Education; 1984.
[32] Choi BK, Kang D. Modeling and simulation of discrete event
systems.. USA: John Wiley and Sons; 2013.
[33] Muller D. AutoModTM-providing simulation solutions for over 25 years. In:
Proceedings of the 2011 winter simulation conference (WSC). IEEE; 2011. p.
39–51.
[34] Concannon KH, Hunter KI, Tremble JM. Dynamic scheduling II: SIMUL8-
planner simulation-based planning and scheduling. In: Proceedings of the
35th conference on winter simulation: driving innovation. Winter simulation
conference. p. 1488–93.
[35] Runciman N, Vagenas N, Corkal T. Simulation of haulage truck loading
techniques in an underground mine using WITNESS. Simulation 1997;68
(5):291–9.
[36] Kamrani M, Abadi SMHE, Golroudbary SR. Traffic simulation of two adjacent
unsignalized T-junctions during rush hours using Arena software. Simul Model
Pract Theory 2014;49:167–79.
[37] Runge I. Mining economics and strategy. Co (USA): Littleton; 2005.